Literature search results
Based on the screening strategy, 426 studies were considered, of which 87 were from Chinese databases. Ninety-seven studies that were duplicates were excluded by the document management software NoteExpress (version 3.2). After screening the titles and abstracts, 254 studies were removed. The full texts of the remaining 75 papers were then screened by the selection criteria. A total of 35 eligible articles were included in the meta-analysis after screening, according to the inclusion and exclusion criteria. The details of the selection process are shown in Fig. 1. Due to the study discussed in literature 7 being conducted across two time periods (1997 and 2003), we treated the data from the two periods as two different studies (literature 7-1, 7-2) in the meta-analysis. Similarly, the data for the study discussed in literature 33 were acquired from 6 areas (Gansu, Ningxia, Qinghai, Sichuan, Tibet, and Xinjiang), therefore literature 33 was regarded as 6 separate studies (literature33-1, 33-2, 33-3, 33-4, 33-5, 33-6).
Basic information of included literatures
Of the 41 selected studies (the splitting up of literature 7 and literature 33 were included), 11 (27%) were in English and 30 (73%) in Chinese (Table 1). The periods of time during which these studies were carried out and published were from 1991 to 2016 and from 1992 to 2019, respectively. All surveys were cross-sectional studies and were conducted in 6 provinces in China: 9 in Gansu, 3 in Ningxia, 8 in Qinghai, 8 in Sichuan, 9 in Tibet, and 4 in Xinjiang. A total of 2,032,811 subjects were included, of which 7522 were HAE patients. Table 1 shows the general features of the included studies.
Table 1 List of included articles in the meta-analysis
Literature quality evaluation
The quality of the cross-sectional studies was evaluated by the adjusted AHRQ checklist (9 items). The results (Table 1) showed that there were 7 low-quality studies (17%), 27 medium-quality studies (66%), and 7 high-quality studies (17%). The details of the quality evaluation were shown in Additional file 2.
Meta-analysis of the prevalence of HAE in China
Substantial heterogeneity was observed among the included studies (I2=100%, P<0.01, Fig. 2), which is common in most meta-analyses studying prevalence. Therefore, we used a random-effects model to calculate the pooled prevalence. The result indicated that the pooled prevalence of HAE in China was 0.96% (95% confidence interval [CI]: 0.71 to 1.25%) (Table 2). The funnel plot was shown in the Additional file 3 and the Egger’s test did not indicate any publication bias (P=0.06). The sensitivity analysis indicated that despite excluding studies, the pooled prevalence remained stable (Additional file 4).
All the studies were divided into 6 groups based on the geographical areas for the subgroup analysis. The prevalences of HAE in these 6 areas were statistically significant (P<0.01, Table 2). The highest prevalence of HAE was in Sichuan (2.03%, 95% CI: 1.30 to 2.92%), while the lowest was in Xinjiang (0.013%, 95% CI: 0.001 to 0.036%). There were not enough studies based in Xinjiang (4 studies) and Ningxia (3 studies), therefore it is possible the pooled prevalences of these two areas may not be accurate.
Table 2 Meta-analysis of HAE prevalence in China
Factors potentially influencing HAE prevalence
Three factors potentially affecting HAE prevalence were considered in our meta-analysis: sex, age, and occupation. Due to the small number of articles addressing “age” and “occupation” (<10 articles), publication bias was only evaluated for “sex.” Heterogeneity was indicated within all groups (sex, age, and occupation) (Additional file 5), with statistically significant differences observed (P<0.01) (Table 2).
Of the included studies, 11 articles (1-3, 8-11, 13, 14, 18, 22) reported data on sex in their study populations. The meta-analysis showed that the prevalence of women with HAE (3.47%, 95% CI: 2.69 to 4.34%) was higher than that of men with HAE (2.14%, 95% CI: 1.55 to 2.82%) (Table 2). The overall pooled odds ratio (OR) for women with HAE was 1.60 (95% CI: 1.35 to 1.91, P<0.01). After excluding literature 1 from the sensitivity analysis, we found that the heterogeneity reduced (I2=27%, P=0.20), and the adjusted OR was 1.50 (95% CI: 1.30 to 1.73, P<0.01), as shown in Additional file 6. The funnel plot (Additional file 3) and Egger’s test did not indicate the existence of publication bias.
Five articles (1, 13, 14, 16, 22) had data on the age of the subjects. The prevalence of HAE was higher in those whose age was ≥30 years old (6.41%, 95% CI: 3.50 to 10.12%) than those whose age was <30 years old (1.64%, 95% CI: 0.36 to 3.83%) (Table 2). Being in the ≥30 years old age group was associated with an increase in HAE prevalence (OR=4.72, 95% CI: 2.29 to 9.75, P<0.01). The sensitivity analysis (Additional file 7) indicated that the OR was relatively stable when any study was excluded.
The occupations of the subjects were investigated in 4 articles (3, 9, 18, 22). The analysis showed that HAE prevalence was higher in farmers and/or herdsmen (2.90%, 95% CI: 1.82 to 4.21%) than in people in other occupations (1.20%, 95% CI: 0.63 to 1.94%), and the OR was 2.54 (95% CI: 1.60 to 4.02, P<0.01). The sensitivity analysis indicated that heterogeneity decreased (I2=50%, P=0.14) when literature 22 was excluded; the adjusted OR was 3.01 (95% CI: 2.02 to 4.48, P<0.01) (Additional file 8).
Trend in the variation of HAE prevalence in China
Fig. 3 shows the trend in the variation of HAE over time. The overall HAE prevalence has remained low since 2005—despite its resurgence in prevalence from 2005–2010, it remained stable after 2010. Additionally, the meta-regression analysis indicated that there was negative correlation between the time period during which the studies were conducted and the HAE prevalence in China (R2=38.11%, P < 0.01) (Fig. 4).